ILIAS  release_5-1 Revision 5.0.0-5477-g43f3e3fab5f
EntityParser.php
Go to the documentation of this file.
1<?php
2
3// if want to implement error collecting here, we'll need to use some sort
4// of global data (probably trigger_error) because it's impossible to pass
5// $config or $context to the callback functions.
6
11{
12
17 protected $_entity_lookup;
18
24 '/&(?:[#]x([a-fA-F0-9]+)|[#]0*(\d+)|([A-Za-z_:][A-Za-z0-9.\-_:]*));?/';
25 // 1. hex 2. dec 3. string (XML style)
26
32 array(
33 34 => '"',
34 38 => '&',
35 39 => "'",
36 60 => '<',
37 62 => '>'
38 );
39
45 array(
46 'quot' => 34,
47 'amp' => 38,
48 'lt' => 60,
49 'gt' => 62
50 );
51
60 public function substituteNonSpecialEntities($string)
61 {
62 // it will try to detect missing semicolons, but don't rely on it
63 return preg_replace_callback(
64 $this->_substituteEntitiesRegex,
65 array($this, 'nonSpecialEntityCallback'),
66 $string
67 );
68 }
69
79 protected function nonSpecialEntityCallback($matches)
80 {
81 // replaces all but big five
82 $entity = $matches[0];
83 $is_num = (@$matches[0][1] === '#');
84 if ($is_num) {
85 $is_hex = (@$entity[2] === 'x');
86 $code = $is_hex ? hexdec($matches[1]) : (int) $matches[2];
87 // abort for special characters
88 if (isset($this->_special_dec2str[$code])) {
89 return $entity;
90 }
92 } else {
93 if (isset($this->_special_ent2dec[$matches[3]])) {
94 return $entity;
95 }
96 if (!$this->_entity_lookup) {
97 $this->_entity_lookup = HTMLPurifier_EntityLookup::instance();
98 }
99 if (isset($this->_entity_lookup->table[$matches[3]])) {
100 return $this->_entity_lookup->table[$matches[3]];
101 } else {
102 return $entity;
103 }
104 }
105 }
106
116 public function substituteSpecialEntities($string)
117 {
118 return preg_replace_callback(
119 $this->_substituteEntitiesRegex,
120 array($this, 'specialEntityCallback'),
121 $string
122 );
123 }
124
135 protected function specialEntityCallback($matches)
136 {
137 $entity = $matches[0];
138 $is_num = (@$matches[0][1] === '#');
139 if ($is_num) {
140 $is_hex = (@$entity[2] === 'x');
141 $int = $is_hex ? hexdec($matches[1]) : (int) $matches[2];
142 return isset($this->_special_dec2str[$int]) ?
143 $this->_special_dec2str[$int] :
144 $entity;
145 } else {
146 return isset($this->_special_ent2dec[$matches[3]]) ?
147 $this->_special_ent2dec[$matches[3]] :
148 $entity;
149 }
150 }
151}
152
153// vim: et sw=4 sts=4
static unichr($code)
Translates a Unicode codepoint into its corresponding UTF-8 character.
Definition: Encoder.php:309
static instance($prototype=false)
Retrieves sole instance of the object.
Handles referencing and derefencing character entities.
$_special_dec2str
Decimal to parsed string conversion table for special entities.
$_substituteEntitiesRegex
Callback regex string for parsing entities.
$_special_ent2dec
Stripped entity names to decimal conversion table for special entities.
substituteNonSpecialEntities($string)
Substitutes non-special entities with their parsed equivalents.
specialEntityCallback($matches)
Callback function for substituteSpecialEntities() that does the work.
$_entity_lookup
Reference to entity lookup table.
nonSpecialEntityCallback($matches)
Callback function for substituteNonSpecialEntities() that does the work.
substituteSpecialEntities($string)
Substitutes only special entities with their parsed equivalents.
$code
Definition: example_050.php:99